Data cleaning method of resource management knowledge graph

Knowledge graph is an important part of artificial intelligence. In terms of resource management, knowledge graph can label and aggregate resources, so as to complete intelligent matching and rapid retrieval of resources. However, when we constructed knowledge graph of resource management, there were some phenomena, such as redundancy of entities, isolated entities and so on. So in the face of these problems, we propose a data cleaning method.Its implementation has four stages: In stage 1, we use NLP technology to detect and classify problematic data. In stage 2, for the single problematic entities, we make pattern statistics and design a relationship supplement method based on relationship reasoning and a new entities alignment method. In stage 3, we recursively supplement and align knowledge fragment sets formed by interconnection of multiple problematic entities. In stage 4, for these isolate knowledge fragment sets, we propose an alignment and fusion method based on multiple entity relationships. Finally, we conducted experiments on our knowledge graph to prove the accuracy and effectiveness of our method.